> 1. I should start udhcpc from an init.d script at boot time with the > "-s" switch to point it to a script that will be called each time an > event occurs that udhcpc must handle
Yes, exactly. > 2. Under normal circumstances, I should just let udhcpc keep running > in the background and it will handle events on its own Yes. > However, how should I tell udhcp that I want it to perform a task such > as releasing a bail? Should I call it through eg. "udhcpc release", > send a signal, or go through the runsvdir/runsv? If you don't need the lease anymore, just kill udhcpc, i.e. send it a SIGTERM. Restart udhcpc if you need another lease. That's all. > Incidently, I'm not clear about how udhcpc and runsvdir/runsv work > together > > BTW, how should I launch udhcpc from the init.d script? Is it required > to use the "-f" switch? If you have no experience with supervision suites and want to just run udhcpc from an init.d script, forget what I said about runsvdir and runsv, and write your standard run-of-the-mill sysv script with start) stop) restart) and status) stuff, launching udhcpc *without* the -f switch, and making sure it stores its pid into some file (typically /var/run/udhcpc-eth0.pid) so you can send it signals. runsvdir/runsv is a whole other way of managing services (a "supervision tree") with a different mindset than sysv scripts. It's a lot better, but requires getting used to. If you're struggling with making udhcpc work, don't attempt to switch to the supervision tree model at the same time. First get a configuration that works for you and that you understand well; when you have udhcpc up and running the way you want to, you can come back and learn how runsvdir/runsv works. -- Laurent _______________________________________________ busybox mailing list [email protected] http://lists.busybox.net/mailman/listinfo/busybox
